Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm

Pull kvm fixes from Paolo Bonzini:
 "x86 and selftests fixes.

  x86:

   - When emulating a guest TLB flush for a nested guest, flush vpid01,
     not vpid02, if L2 is active but VPID is disabled in vmcs12, i.e. if
     L2 and L1 are sharing VPID '0' (from L1's perspective).

   - Fix a bug in the SNP initialization flow where KVM would return '0'
     to userspace instead of -errno on failure.

   - Move the Intel PT virtualization (i.e. outputting host trace to
     host buffer and guest trace to guest buffer) behind CONFIG_BROKEN.

   - Fix memory leak on failure of KVM_SEV_SNP_LAUNCH_START

   - Fix a bug where KVM fails to inject an interrupt from the IRR after
     KVM_SET_LAPIC.

  Selftests:

   - Increase the timeout for the memslot performance selftest to avoid
     false failures on arm64 and nested x86 platforms.

   - Fix a goof in the guest_memfd selftest where a for-loop initialized
     a bit mask to zero instead of BIT(0).

   - Disable strict aliasing when building KVM selftests to prevent the
     compiler from treating things like "u64 *" to "uint64_t *" cases as
     undefined behavior, which can lead to nasty, hard to debug
     failures.

   - Force -march=x86-64-v2 for KVM x86 selftests if and only if the
     uarch is supported by the compiler.

   - Fix broken compilation of kvm selftests after a header sync in
     tools/"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: VMX: Bury Intel PT virtualization (guest/host mode) behind CONFIG_BROKEN
  KVM: x86: Unconditionally set irr_pending when updating APICv state
  kvm: svm: Fix gctx page leak on invalid inputs
  KVM: selftests: use X86_MEMTYPE_WB instead of VMX_BASIC_MEM_TYPE_WB
  KVM: SVM: Propagate error from snp_guest_req_init() to userspace
  KVM: nVMX: Treat vpid01 as current if L2 is active, but with VPID disabled
  KVM: selftests: Don't force -march=x86-64-v2 if it's unsupported
  KVM: selftests: Disable strict aliasing
  KVM: selftests: fix unintentional noop test in guest_memfd_test.c
  KVM: selftests: memslot_perf_test: increase guest sync timeout

Author: torvalds

arch/x86/kvm/lapic.c

@@ -2629,19 +2629,26 @@ void kvm_apic_update_apicv(struct kvm_vcpu *vcpu)
 {
 	struct kvm_lapic *apic = vcpu->arch.apic;
 
-	if (apic->apicv_active) {
-		/* irr_pending is always true when apicv is activated. */
-		apic->irr_pending = true;
+	/*
+	 * When APICv is enabled, KVM must always search the IRR for a pending
+	 * IRQ, as other vCPUs and devices can set IRR bits even if the vCPU
+	 * isn't running.  If APICv is disabled, KVM _should_ search the IRR
+	 * for a pending IRQ.  But KVM currently doesn't ensure *all* hardware,
+	 * e.g. CPUs and IOMMUs, has seen the change in state, i.e. searching
+	 * the IRR at this time could race with IRQ delivery from hardware that
+	 * still sees APICv as being enabled.
+	 *
+	 * FIXME: Ensure other vCPUs and devices observe the change in APICv
+	 *        state prior to updating KVM's metadata caches, so that KVM
+	 *        can safely search the IRR and set irr_pending accordingly.
+	 */
+	apic->irr_pending = true;
+
+	if (apic->apicv_active)
 		apic->isr_count = 1;
-	} else {
-		/*
-		 * Don't clear irr_pending, searching the IRR can race with
-		 * updates from the CPU as APICv is still active from hardware's
-		 * perspective.  The flag will be cleared as appropriate when
-		 * KVM injects the interrupt.
-		 */
+	else
 		apic->isr_count = count_vectors(apic->regs + APIC_ISR);
-	}
+
 	apic->highest_isr_cache = -1;
 }
 

arch/x86/kvm/svm/sev.c

@@ -450,8 +450,11 @@ static int __sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp,
 		goto e_free;
 
 	/* This needs to happen after SEV/SNP firmware initialization. */
-	if (vm_type == KVM_X86_SNP_VM && snp_guest_req_init(kvm))
-		goto e_free;
+	if (vm_type == KVM_X86_SNP_VM) {
+		ret = snp_guest_req_init(kvm);
+		if (ret)
+			goto e_free;
+	}
 
 	INIT_LIST_HEAD(&sev->regions_list);
 	INIT_LIST_HEAD(&sev->mirror_vms);
@@ -2212,10 +2215,6 @@ static int snp_launch_start(struct kvm *kvm, struct kvm_sev_cmd *argp)
 	if (sev->snp_context)
 		return -EINVAL;
 
-	sev->snp_context = snp_context_create(kvm, argp);
-	if (!sev->snp_context)
-		return -ENOTTY;
-
 	if (params.flags)
 		return -EINVAL;
 
@@ -2230,6 +2229,10 @@ static int snp_launch_start(struct kvm *kvm, struct kvm_sev_cmd *argp)
 	if (params.policy & SNP_POLICY_MASK_SINGLE_SOCKET)
 		return -EINVAL;
 
+	sev->snp_context = snp_context_create(kvm, argp);
+	if (!sev->snp_context)
+		return -ENOTTY;
+
 	start.gctx_paddr = __psp_pa(sev->snp_context);
 	start.policy = params.policy;
 	memcpy(start.gosvw, params.gosvw, sizeof(params.gosvw));

arch/x86/kvm/vmx/nested.c

@@ -1197,11 +1197,14 @@ static void nested_vmx_transition_tlb_flush(struct kvm_vcpu *vcpu,
 	kvm_hv_nested_transtion_tlb_flush(vcpu, enable_ept);
 
 	/*
-	 * If vmcs12 doesn't use VPID, L1 expects linear and combined mappings
-	 * for *all* contexts to be flushed on VM-Enter/VM-Exit, i.e. it's a
-	 * full TLB flush from the guest's perspective.  This is required even
-	 * if VPID is disabled in the host as KVM may need to synchronize the
-	 * MMU in response to the guest TLB flush.
+	 * If VPID is disabled, then guest TLB accesses use VPID=0, i.e. the
+	 * same VPID as the host, and so architecturally, linear and combined
+	 * mappings for VPID=0 must be flushed at VM-Enter and VM-Exit.  KVM
+	 * emulates L2 sharing L1's VPID=0 by using vpid01 while running L2,
+	 * and so KVM must also emulate TLB flush of VPID=0, i.e. vpid01.  This
+	 * is required if VPID is disabled in KVM, as a TLB flush (there are no
+	 * VPIDs) still occurs from L1's perspective, and KVM may need to
+	 * synchronize the MMU in response to the guest TLB flush.
 	 *
 	 * Note, using TLB_FLUSH_GUEST is correct even if nested EPT is in use.
 	 * EPT is a special snowflake, as guest-physical mappings aren't
@@ -2315,6 +2318,17 @@ static void prepare_vmcs02_early_rare(struct vcpu_vmx *vmx,
 
 	vmcs_write64(VMCS_LINK_POINTER, INVALID_GPA);
 
+	/*
+	 * If VPID is disabled, then guest TLB accesses use VPID=0, i.e. the
+	 * same VPID as the host.  Emulate this behavior by using vpid01 for L2
+	 * if VPID is disabled in vmcs12.  Note, if VPID is disabled, VM-Enter
+	 * and VM-Exit are architecturally required to flush VPID=0, but *only*
+	 * VPID=0.  I.e. using vpid02 would be ok (so long as KVM emulates the
+	 * required flushes), but doing so would cause KVM to over-flush.  E.g.
+	 * if L1 runs L2 X with VPID12=1, then runs L2 Y with VPID12 disabled,
+	 * and then runs L2 X again, then KVM can and should retain TLB entries
+	 * for VPID12=1.
+	 */
 	if (enable_vpid) {
 		if (nested_cpu_has_vpid(vmcs12) && vmx->nested.vpid02)
 			vmcs_write16(VIRTUAL_PROCESSOR_ID, vmx->nested.vpid02);
@@ -5950,6 +5964,12 @@ static int handle_invvpid(struct kvm_vcpu *vcpu)
 		return nested_vmx_fail(vcpu,
 			VMXERR_INVALID_OPERAND_TO_INVEPT_INVVPID);
 
+	/*
+	 * Always flush the effective vpid02, i.e. never flush the current VPID
+	 * and never explicitly flush vpid01.  INVVPID targets a VPID, not a
+	 * VMCS, and so whether or not the current vmcs12 has VPID enabled is
+	 * irrelevant (and there may not be a loaded vmcs12).
+	 */
 	vpid02 = nested_get_vpid02(vcpu);
 	switch (type) {
 	case VMX_VPID_EXTENT_INDIVIDUAL_ADDR:

arch/x86/kvm/vmx/vmx.c

@@ -217,9 +217,11 @@ module_param(ple_window_shrink, uint, 0444);
 static unsigned int ple_window_max        = KVM_VMX_DEFAULT_PLE_WINDOW_MAX;
 module_param(ple_window_max, uint, 0444);
 
-/* Default is SYSTEM mode, 1 for host-guest mode */
+/* Default is SYSTEM mode, 1 for host-guest mode (which is BROKEN) */
 int __read_mostly pt_mode = PT_MODE_SYSTEM;
+#ifdef CONFIG_BROKEN
 module_param(pt_mode, int, S_IRUGO);
+#endif
 
 struct x86_pmu_lbr __ro_after_init vmx_lbr_caps;
 
@@ -3216,7 +3218,7 @@ void vmx_flush_tlb_all(struct kvm_vcpu *vcpu)
 
 static inline int vmx_get_current_vpid(struct kvm_vcpu *vcpu)
 {
-	if (is_guest_mode(vcpu))
+	if (is_guest_mode(vcpu) && nested_cpu_has_vpid(get_vmcs12(vcpu)))
 		return nested_get_vpid02(vcpu);
 	return to_vmx(vcpu)->vpid;
 }

tools/testing/selftests/kvm/Makefile

@@ -241,16 +241,18 @@ CFLAGS += -Wall -Wstrict-prototypes -Wuninitialized -O2 -g -std=gnu99 \
 	-Wno-gnu-variable-sized-type-not-at-end -MD -MP -DCONFIG_64BIT \
 	-fno-builtin-memcmp -fno-builtin-memcpy \
 	-fno-builtin-memset -fno-builtin-strnlen \
-	-fno-stack-protector -fno-PIE -I$(LINUX_TOOL_INCLUDE) \
-	-I$(LINUX_TOOL_ARCH_INCLUDE) -I$(LINUX_HDR_PATH) -Iinclude \
-	-I$(<D) -Iinclude/$(ARCH_DIR) -I ../rseq -I.. $(EXTRA_CFLAGS) \
-	$(KHDR_INCLUDES)
+	-fno-stack-protector -fno-PIE -fno-strict-aliasing \
+	-I$(LINUX_TOOL_INCLUDE) -I$(LINUX_TOOL_ARCH_INCLUDE) \
+	-I$(LINUX_HDR_PATH) -Iinclude -I$(<D) -Iinclude/$(ARCH_DIR) \
+	-I ../rseq -I.. $(EXTRA_CFLAGS) $(KHDR_INCLUDES)
 ifeq ($(ARCH),s390)
 	CFLAGS += -march=z10
 endif
 ifeq ($(ARCH),x86)
+ifeq ($(shell echo "void foo(void) { }" | $(CC) -march=x86-64-v2 -x c - -c -o /dev/null 2>/dev/null; echo "$$?"),0)
 	CFLAGS += -march=x86-64-v2
 endif
+endif
 ifeq ($(ARCH),arm64)
 tools_dir := $(top_srcdir)/tools
 arm64_tools_dir := $(tools_dir)/arch/arm64/tools/

tools/testing/selftests/kvm/guest_memfd_test.c

@@ -134,7 +134,7 @@ static void test_create_guest_memfd_invalid(struct kvm_vm *vm)
 			    size);
 	}
 
-	for (flag = 0; flag; flag <<= 1) {
+	for (flag = BIT(0); flag; flag <<= 1) {
 		fd = __vm_create_guest_memfd(vm, page_size, flag);
 		TEST_ASSERT(fd == -1 && errno == EINVAL,
 			    "guest_memfd() with flag '0x%lx' should fail with EINVAL",

tools/testing/selftests/kvm/lib/x86_64/vmx.c

@@ -200,7 +200,7 @@ static inline void init_vmcs_control_fields(struct vmx_pages *vmx)
 	if (vmx->eptp_gpa) {
 		uint64_t ept_paddr;
 		struct eptPageTablePointer eptp = {
-			.memory_type = VMX_BASIC_MEM_TYPE_WB,
+			.memory_type = X86_MEMTYPE_WB,
 			.page_walk_length = 3, /* + 1 */
 			.ad_enabled = ept_vpid_cap_supported(VMX_EPT_VPID_CAP_AD_BITS),
 			.address = vmx->eptp_gpa >> PAGE_SHIFT_4K,

tools/testing/selftests/kvm/memslot_perf_test.c

@@ -417,7 +417,7 @@ static bool _guest_should_exit(void)
  */
 static noinline void host_perform_sync(struct sync_area *sync)
 {
-	alarm(2);
+	alarm(10);
 
 	atomic_store_explicit(&sync->sync_flag, true, memory_order_release);
 	while (atomic_load_explicit(&sync->sync_flag, memory_order_acquire))